Sun to use Nvidia chipset for 8-way servers?

G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

http://www.theinquirer.net/?article=16403

Looks like Sun is playing off Nvidia against Serverworks to get the cheapest
8-way chipset they can. Welcome to the commodity super-server era, brought
to you by AMD.

Yousuf Khan

--
Humans: contact me at ykhan at rogers dot com
Spambots: just reply to this email address ;-)
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

Yousuf Khan wrote:
> http://www.theinquirer.net/?article=16403
>
> Looks like Sun is playing off Nvidia against Serverworks to get the cheapest
> 8-way chipset they can. Welcome to the commodity super-server era, brought
> to you by AMD.
>

Hmmm. I see that I can't go on ignoring Opteron forever. :).

The eight-way setups aren't glueless, as was originally advertised, or,
by a "chipset," do they mean just a South bridge? I'm so confused. As
if that weren't enough, in one of the articles linked to at the bottom,

http://www.theinquirer.net/?article=16389,

there is the mysterious statement

"Instead, according to the same sources, Tyan is likely to introduce an
eight way design by the end of the year which will wait for another
chipset, use PCI Express, and also support Infiniband PCI Express too."

*What* is "Infiniband PCI Express"? I suppose I could try to make some
progress on trying to understand what all this is about, but, in
observance of the 50th anniversary of Turing's death, I am spending the
day with Boolean Rings and Lattices.

RM
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

Robert Myers <rmyers1400@comcast.net> wrote:
> Yousuf Khan wrote:
>> http://www.theinquirer.net/?article=16403
>>
>> Looks like Sun is playing off Nvidia against Serverworks to get the
>> cheapest 8-way chipset they can. Welcome to the commodity
>> super-server era, brought to you by AMD.
>>
>
> Hmmm. I see that I can't go on ignoring Opteron forever. :).
>
> The eight-way setups aren't glueless, as was originally advertised,
> or, by a "chipset," do they mean just a South bridge? I'm so
> confused. As if that weren't enough, in one of the articles linked
> to at the bottom,

Probably that's all they mean by chipset these days, just a southbridge.

> *What* is "Infiniband PCI Express"? I suppose I could try to make
> some progress on trying to understand what all this is about, but, in
> observance of the 50th anniversary of Turing's death, I am spending
> the day with Boolean Rings and Lattices.

I was hoping you'd know what it was.

Yousuf Khan
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

Yousuf Khan wrote:

> Robert Myers <rmyers1400@comcast.net> wrote:
>

<snip>

>
>>*What* is "Infiniband PCI Express"? I suppose I could try to make
>>some progress on trying to understand what all this is about, but, in
>>observance of the 50th anniversary of Turing's death, I am spending
>>the day with Boolean Rings and Lattices.
>
>
> I was hoping you'd know what it was.
>

It's a reasonable guess the writer wanted to convey an architecture that
would mimic Intel's layout, which will soon be Infiniband as the server
interconnect of choice (for the moment, at least) outside the box
connected to pci-express inside the box.

In that view, "Infiniband PCI Express" (I squirm every time I type it)
would be in contrast to a straight Hypertransport-Infiniband HCA, of
which Octicabay would be the only potential supplier I know of, if they
are, indeed, in a position to supply such an animal to others.

RM